Multiple Infections with Cardinium and Two Strains of Wolbachia in The Spider Mite Tetranychus phaselus Ehara: Revealing New Forces Driving the Spread of Wolbachia

Cytoplasmic incompatibility (CI) has been proposed as a major mechanism by which certain strains of Wolbachia to invade and persist in host populations. However, mechanisms that underlie the invasion and persistence of non-CI strains are less well understood. Here, we established a spider mite Tetranychus phaselus population multiply infected by Cardinium as well as two distinct lineages of Wolbachia, designated wCon and wOri, to study the forces driving the spread of the non-CI strain of Wolbachia wOri. Interestingly, we found that wOri provided a longevity advantage to its female hosts under ideal conditions, making wOri stay longer in this population, and then being transmitted to more offspring. Furthermore, the lifespan of uninfected females was reduced when mated with multiple-infected males. As a result, the uninfected population is attenuated by the multiple-infected males. Thus, we infer that the host age effects of multiple infection may represent sufficient forces driving the spread of wOri through the host population.


Introduction
Wolbachia are obligate intracellular rickettsia-like bacteria that occur in numerous invertebrates. Infection with Wolbachia is found frequently in insects [1], mites [2,3], spiders [4,5], crustaceans [6], and nematodes [7]. Another maternally inherited symbiont Cardinium from the Bacteroidetes group has been found in four insect orders and approximately 6-7% of arthropods species that have been examined [8,9]. These intracellular bacteria have recently received much attention for their various intriguing reproductive abnormalities that account for their success, including parthenogenesis, feminization, male-killing, and cytoplasmic incompatibility (CI). CI is the developmental arrest of insect embryos that result when females are mated with males that have a different infection status.
Several reports have noted that CI, the most common of the reproductive abnormalities, imparts the ability to Wolbachia to invade host populations [10,11,12]. This effect of CI has been quantified and modeled in several arthropod species including Drosophila simulans [11] and Aedes albopictus [13,14]. Once enough individuals are infected, any disadvantages associated with infection are outweighed by the presence of CI-induced mortality. In other words, the more common the infection is, the more likely it is that uninfected females will encounter infected males and hence suffer reproductive losses. While CI has been proposed as a major mechanism allowing certain strains of Wolbachia to invade and persist in host populations, mechanisms that underlie the invasion and persistence of non-CI strains are less well understood. It is surprising, therefore, that spread mechanism of Wolbachia symbioses has not all been identified yet, especially in view of the fact that several Wolbachia strains have invaded natural populations even though they induce little to no CI [15]. Several reports show that the protection of their hosts from viruses and other pathogens has in recent years been revealed to be a major driving force in Wolbachia [16,17]. It is hypothesized that these strains may provide an as yet undetermined fitness benefit to their hosts. To our knowledge, this hypothesis has not yet been tested.
The spider mite Tetranychus phaselus is a serious agricultural pest that damages many crops including cotton, bean, cucumber, eggplant, pepper, tomato, cucurbits, papaya, passion fruit. Infections of both Wolbachia and Cardinium in the same host are known in a number of species in the Acari and the Hymenoptera [8,9,18,19,20]. Moreover, a number of studies have reported that multiple infections of Wolbachia occurred commonly in several species, Drosophila simulans [21,22], Aedes albopictus [23], Nasonia vitripennis [24,25], Callosobruchus chinensis [26,27], the Drosophila parasitoid Leptopilina heterotoma [28] and Tetranychus cinnabarinus (T. urticae red form) [29]. Recently, we found a multiple infection pattern in the spider mite T. phaselus. Wolbachia-superinfection (that is co-infection with two or more strains), and Cardinium-infection occur naturally in the same host. To our knowledge, this is the first case of multiple infection with Cardinium and two strains of Wolbachia in arthropods.
The aim of the present study was to discover the mechanism by which wOri is spread in co-existence with wCon and Cardinium in T. phaselus. To this end, we established three lines of T. phaselus, an uninfected line (Wolbachia-free and Cardinium-free), a doubleinfected line (infected by wCon and Cardinium), and a multipleinfected line (infected by wCon, wOri and Cardinium).We examined the levels of CI induced in T. phaselus by double-infection and multiple-infection, and compared the longevity, egg hatch, fecundity and bacteria densities in the three lines of T. phaselus. Our results reveal that wOri has effects on host age that may act as new, non-CI forces for the spread of a Wolbachia strain.

Preparation of Spider Mite Lines
The spider mite Tetranychus phaselus was collected from soybean [Glycine max (L.) Merr.] leaves in Cixi, Zhejiang Province, east China in July 2009. Mites were reared on a leaf of the common bean (Phaseolus vulgaris L.) placed on a water-saturated sponge mat in Petri dishes (dia. 9) at 2561uC, 60% r.h. and under L16-D8 conditions. In order to cross infected and uninfected individuals, 100% double-infected, 100% multiple-infected and 100% uninfected lines were prepared for the population. One female from the teleiochrysalis stage was allowed to lay eggs without being crossed with males. The eggs were reared until adulthood (males). After the males reached sexual maturity, they were backcrossed with the mother. Then, the female adults were transferred to new leaf discs and were allowed to lay eggs for 3-5 days. A female was checked for Wolbachia and Cardinium infection status by PCR amplification. The eggs were separately reared on new leaf discs depending on the infection status of the mother. The above process was continued for three to four generations until a 100% infected population was obtained. The eggs of the uninfected mothers were reared to establish the uninfected line.

DNA Extraction and Diagnostic PCR
In order to exclude the influence of different body sizes of female and male mites, DNA was extracted by homogenizing a single adult mite in a 15 ml (male) or 25 ml (female) mixture of STE buffer (100 mM NaCl, 10 mM Tris-HCl, 1 mM EDTA, pH 8.0) and proteinase K (10 mg/ml, 2 ml) in a 1.5 ml Eppendorf tube. The mixture was incubated at 37uC for 30 min and then 95uC for 5 min. The least bacteria DNA is lost with this method of DNA extraction.
For samples failing to amplify using Wolbachia and Cardiniumspecific primers, primers par COI-forward and COI-reverse [30] were used to amplify mitochondria DNA as a positive control for template DNA quality. Approximately 7 ml of the PCR products were electrophoresed in a 1.0% agarose gel in TBE/EtBr for 40 min at 60 mA, and then photographed on a UV transilluminator. Amplified fragments were purified using a Gel Extraction Mini kit (Watson, Shanghai, China). Then, the distinct single-band amplicons were cloned into pGEM T-Easy Vector (Promega, Madison, WI, USA) and the positive clones were screened and sequenced directly. The sequence data helps us to confirm that the PCR positive individuals are infected with wOri, wCon and Cardinium.

Cross Experiments
To reveal how the different Wolbachia strains and Cardinium influenced CI modification and rescue, crossing experiments were performed (Table 1). Female teleiochrysalids, the last developmental stage before adult emergence, were placed with two males on the same leaf disk. We used 1-day-old virgin males produced as a cohort by groups of females isolated as teliochrysalids. This procedure was designed to avoid the potential decrease of the Wolbachia and Cardinium effects due to male ageing or repeated consecutive mating. Males were discarded 2 days after the females' eclosion, and mated females were allowed to oviposit for 5 days. Eggs on leaf discs were checked daily to determine the hatchability, sex ratio (% daughters) and mortality of offspring. Fecundity was estimated as the total number of eggs laid in the first 5 days. Data were analyzed with one-way analysis of variance (ANOVA), and the means were compared using the Tukey-HSD test (SPSS 17.0). To normalize the data, log transformation was used for the number of eggs laid per female, and arcsine square root transformation was used for egg hatchability, sex ratio and mortality. Eight super-infected males of 1-, 4-, 8-and 12-day-old were collected. DNA of single mites was extracted using the above method. Three replicates were run and averaged for each DNA sample. Negative controls were included in all amplification reactions.

Bacteria Density Measurement
Standard curves were plotted using a 10-fold dilution series consisting of 10 27 to 10 23 dilutions of the DNA standards prepared from plasmid DNA. The quality and concentration of all purified standard DNA were measured by OD absorbance at 260 nm. The number of molecules in all samples was determined from the threshold cycles in the PCR based on the standard curve. Statistical analysis was performed using the Mann-Whitney U-test.

Survival Assessment
Differences in host longevity were observed in comparisons of the different infection types. We measured age-specific survival of the females that are uninfected, single-infected and doubleinfected. For the infected females, we compared the longevity difference between single-infected and double-infected virgins. Meanwhile, the longevity of uninfected females was also measured. They are divided into four groups, according to their different mating patterns, including virgins as well as those mated with uninfected, single-infected and double-infected males. In order to test the longevity of virgins, 50 teleiochrysalids (the last develop-mental stage before eclosion) were placed on the different leaves according to the infection patterns. The new adults emerging within 24 hours were tested for longevity. For the longevity of mated females, every newly emerging adult was separately placed with two males with different infection statuses (uninfection, double-infection and multiple-infection) on the same leaf disk. The males were allowed to stay with females for 48 hours, which was designed to avoid the variance in mating times. Leaves were monitored every day, and dead females were removed and counted until all females had died. Survivor curves for individual hosts were compared using the Kaplan-Meier log-rank test (SPSS 17.0).

PCR Detection of wOri and wCon in Spider Mites
Thirteen out of forty mites in a rearing group were positive in PCR assay for Wolbachia infection, among which four were infected only by wCon and nine were superinfected by both wOri and wCon. Moreover, we detected Cardinium 16S rDNA in all theWolbachia positive mites. However, we did not discover any individuals singly infected with Cardinium, wOri or wCon. As a result, we established 3 lines including a 100% uninfected line (U), a 100% double-infected line (infected with wCon and Cardinium, Iwc) and a 100% multiple-infected line (infected with wOri, wCon and Cardinium, Iwwc).

Effects of wOri Effects on Host Reproduction
In order to clarify the effect of wOri on host reproduction, four crosses between Iwc and Iwwc were conducted.
The hatchability of eggs was compared among the four crosses in which the CI of wOri was investigated ( Fig.1. A, B). No significant differences were found between the predicted incompatible cross (Iwc6Iwwc) and the other compatible crosses for any of the investigated traits. We found that males extra infection with wOri (Iwwc) were not able to induce CI with single-infected females (Iwc). Therefore, under this intricate infection status, wOri existed as a non-CI strain in this spider mite population. Meanwhile, CI expression of double-infected (Iwc) and multipleinfected (Iwwc) males was also contrasted. There was no significant difference in hatchability and sex ratio between the two predicted CI crosses (U6Iwc; U6Iwwc), indicating that wOri does neither contribute to induce a higher level of CI nor induce additive effects on CI expression.

Effects of Double-infection and Multiple-infection on Host Reproduction
The two groups of crosses in which CI induced by the doubleinfection and multiple-infection were detected (U6Iwc and U6Iwwc) showed a strong reduction in egg hatchability and a strongly male-biased sex ratio. In the predicted incompatible cross (RU6=Iwc), on average, (26.0263.94)% of all eggs hatched, against 93.5-93.8% in the other crosses ( Fig.1.C). Similarly, the mean egg hatchability between uninfected females and multipleinfected males (RU6=Iwwc) was (23.8464.63)%, significantly lower than that found in compatible crosses (93.2%-96.7%) ( Fig.1.E). The sex ratios of the offspring that did hatch in these two incompatible crosses were significantly lower than the sex ratios from the compatible crosses ( Fig.1.D, F). These crosses displayed a high level of CI expression. All mites used in this research were infected and uninfected naturally, without any antibiotic treatment for exclusion of extra interferential factors. We were unable to find any individuals singly infected with Wolbachia or Cardinium, which made it difficult to evaluate the effects. All that we can say from  The Diverse Forces Driving Spread of Wolbachia PLOS ONE | www.plosone.org these limited results is that the strength of CI induced by Wolbachia and Cardinium together is respectively strong. From the results of bacteria measurements, we found that the density of wCon was close to the density of Cardinium, due to which we infer that both wCon and Cardinium contribute to the strong CI.

Effects of wOri, wCon and Cardinium Infection on Host Fecundity
Fecundity of uninfected, single-infected and super-infected females was compared by analyzing the number of eggs laid in the first 5 days (Table.1). From these crosses we can draw three conclusions: (i) Mating with uninfected males increases the fecundity of double-infected females, which consequently promotes the spread of wCon and Cardinium. (ii) The fecundity of multiple-infected females can be increased by mating with uninfected males, but decreased by mating with double-infected males. It means that double-infected males act as a block to expansion of multiple infection by weakening the fecundity of multiple-infected females. Based on this point, we agree with your opinion that there is a competition between the doubly and triply infected females. (iii) Both the double-infected and multipleinfected males strongly depress the fecundity of their uninfected mates. In particular, multiple-infected males caused an even greater depression of fecundity.

Bacteria Density Measurement
To understand why multiple-infected males were unable to induce a higher level of CI, we examined the infection densities of wOri, wCon and Cardinium in the double-and multiple-infected individuals, in which the CI strength seemed to be related to the bacteria density. Our hunch proved to be correct: wOri was present at a rather low density in the multiple-infected males, being only 0.1%-0. 2% of the densities of wCon and Cardinium (Fig.2).
In order to reveal whether the existence of wOri affects the proliferation of wCon and Cardinium, we compared the wCon and Cardinium densities in double-and multiple-infected females separately (Fig.3).The double-infected females have higher wCon and Cardinium copy numbers than the multiple-infected females when they were 4 days old. For the older females (from 8 days old to 20 days old), infection with wOri enhanced the proliferation of wCon and Cardinium. Furthermore, the dynamic of wCon density during host aging are affected by wOri infection. wCon density increased more rapidly in multiple-infected females (from (0.4060.04)610 8 /ml to (4.9560.14) 610 8 /ml) than in doubleinfected females (from (0.4660.03)610 8 /ml to (3.8760.39) 610 8 / ml) during the first 16 days of their life spans (Fig.3.a). In addition, there are different dynamics of Cardinium density in double-infected females and multiple-infected females. The density of Cardinium in double-infected females was (1.0360.05)610 8 /ml on the first day post eclosion (Fig.3.b). However, from the fourth day to the twentieth day, the density fluctuated between (3.9360.39)610 8 and (4.6460.29)610 8 copies per ml (Fig.3b). During this period, Mann-Whitney tests indicated that the differences between each pair of two ages were not significant (P.0.05). However, in multiple-infected females, Cardinium density rose from (1.0760.11)610 8 /ml to (7.1060.46) 610 8 /ml step by step from the first day to the sixteenth day post eclosion, and then, it fells to (5.4360.52)610 8 /ml on the twentieth day. Mann-Whitney tests indicated that the differences between each pair of two ages were significant (P,0.05).

Survival Assessment
To discover the effects that wOri had on its female hosts, we measured the longevities of uninfected (U), double-infected (Iwc) and multiple-infected (Iwwc) females. Because of the variously uncertain factors in the environment, it is almost impossible to obtain the natural dynamics of the longevity of the hosts on different infection statuses. In order to exclude any influence of differences in environmental conditions, we measured the females on different infection statuses under ideal conditions, including no predation, no food limitation and no mating behavior. On ideal conditions (no predation, no food limitation and no mating behavior), we found that wOri infection prolonged female survival ( Fig.4.a). This means that the longevity of multiple-infected females (25.0861.88 days) was longer than the longevities of double-infected females (18.3561.26 days) and uninfected females (19.6261.65 days; x 2 = 13.43, df = 1, P,0.001; x 2 = 3.894, df = 1, P = 0.048). Consequently, wOri can remain longer in the female hosts, which allows it to be transmitted to more offspring. We regard this phenomenon as a force causing wOri to persist in this population under ideal conditions.
To determine the effect of infection status of males on the longevity of females, we compared the longevities of different groups of uninfected females which are virgins, and those separately mated with uninfected, double-infected and multiple- Copy numbers per ml were determined by quantitative PCR using the wOri wsp gene, wCon wsp gene and 16S rDNA gene (error bars represent 95% bootstrap confidence intervals, n = 8 per each column). a, b and c represent statistic groups (Tukey-HSD test, p,0.05); NS, not significant at the 5% level. doi:10.1371/journal.pone.0054964.g002 infected males. From the survival curves of these four groups of females ( Fig.4.b), three conclusions can be drawn. First, virgin females (line A, 19.6261.65 days) have a longevity advantage over mated females (line B, 12.4961.73 days; line C, 11.2961.28 days; line D, 7.4060.95 days). This means that mating behavior is a direct factor that shortens the females' lifespan. Second, no significant difference was found between the survival curves of females (line B, 12.4961.73 days) mated with uninfected males and females (line C, 11.2961.28 days) mated with double-infected males (x 2 = 0.696, df = 1, P = 0.404). Third, mating with multipleinfected males dramatically shortened the life span of uninfected females (line D, 7.4060.95 days). The difference in longevity  between females mated with uninfected males and females mated with multiple-infected males was significant (x 2 = 5.015, df = 1, P = 0.025). Also, females mated with multiple-infected males have a significantly shorter life span than females mated with doubleinfected males (x 2 = 6.398, df = 1, P = 0.011). These results clearly show that the longevity of uninfected females mated with multipleinfected males was further shortened by not only mating behavior, but also the spermatozoa modified with wOri. The ability of wOri to reduce the longevity of uninfected females by their spermatozoa may act as an additional force to drive the spread of wOri through a mite population. This effect of wOri gives extra strength to a multiple-infection to invade an uninfected population.

Discussion
To our knowledge, this is the first case of multiple infection with Cardinium and two strains of Wolbachia in arthropods. A possible mechanism of the multiple infections in the wild is horizontal transmission. An infection survey of wild spider mite populations in Zhejiang Province shows that Tetranychus truncatus, another important agriculture pest co-existing in this area, is singly infected by wOri (own unpublished data). Probably, wOri in T. phaselus was acquired through horizontal transmission from T. trunctatus. It could also be horizontal transmitted from T. phaselus to T. trunctatus. Another possibility is that both species could have inherited the triple infection from a common ancestor and Cardinium and wCon became lost in T. trunctatus [31,32,33]. However, further study is needed to test this hypothesis.
The results of bacteria density measurements and crossing experiments appear to show there is a reciprocal relationship between wOri and the other bacteria (wCon and Cardinium) in the same host. wCon and Cardinium share their intensive CI with wOri, although wCon and Cardinium are dominant strains over wOri, which compensates for the CI absence of wOri. Even though wOri suppressed the densities of wCon and Cardinium in the four-dayold multiple-infected females because of the limited nutrients in the young hosts, it had favorable effects on the densities of wCon and Cardinium in the older hosts (Fig. 3). During the medium-term of their life span (from the eighth day to the twentieth day), the main oviposition period of the spider mite, wOri significantly facilitated the proliferation of wCon and Cardinium. That means more wCon and Cardinium can be transmitted to offspring in multiple-infected females than in double-infected females. In general, the reciprocal relationship allows the three strains of bacteria to co-exist in the same host.
The powerful spreading capability of Wolbachia has attracted considerable attention as a potential gene driving system. The expression of transgenes could be simplified by exploiting the population invasion ability of Wolbachia. Within this theoretical framework, the capability of Wolbachia to invade and maintain themselves in host populations depends on three main parameters: (i) the strength of CI, (ii) maternal transmission efficiency, and (iii) fitness effects on the host [34].
An empirical study directly examined the transmission efficiency of Wolbachia in spider mites in the laboratory [35], during which a perfect maternal transmission has been found. On this basis, we infer that transmission efficiency of wOri is perfect probably, ensuring its maintenance in the T. phaselus population. However, future experiments are needed to examine our conjecture.
Cytoplasmic incompatibility (CI) is most commonly associated with Wolbachia infections. The intensity of CI is an important parameter that determines the spread and maintenance of Wolbachia infection in the host population [36]. CI results in Wolbachia spread, as witnessed in nature in Drosophila simulans [10] and in a laboratory population of Wolbachia-transinfected A. aegypti [37]. Some strains of Wolbachia induce high CI, which leads to rapid population invasion. For instance, the Wolbachia strain wRi has been shown to spread geographically at a rate of 100 km/year into uninfected D. simulans populations through the action of CI [11].
Wolbachia super-infections (that is co-infection with two or more Wolbachia strains) occur naturally [14,22,38] and have been generated artificially as well [39,40]. These super-infections typically have additive effects, such that a super-infected male is unidirectionally incompatible with both single-infected and uninfected females, whereas super-infected females can rescue infections in both super-infected and single-infected females. In the present study, superinfection with wOri did not induce a higher level of CI, indicating that wOri did not have an additive effect. The bacteria density measurements lend qualified support to the hypothesis that wCon and Cardinium interfere with the ability of wOri to cause CI by suppressing wOri densities. wOri was present in the lowest copy number among those bacteria in males, being only about 0.1%-0.2% of the densities of wCon and Cardinium. Obviously, wCon and Cardinium are dominant species compared with wOri. Similarly, suppression of the density of one symbiont by the other has been found before in natural infections [41,42], and density differences may also have been the reason for the lack of CI modification and rescue in the Wolbachia super-infection created in mosquitoes by Walker et al [40]. Moreover, studies of the interaction of Wolbachia with developing sperm show that an abundance of Wolbachia in the testes is required to induce CI. We hypothesized that the testes of T. phaselus have been occupied by wCon and Cardinium, leaving limited room for wOri. This is a potential mechanism of undetectable CI of wOri, which could be further proved by localizing the symbionts in testes of multipleinfected males [43,44]. In this study, the interactions between wCon and Cardinium could not be investigated because of the absence of lines of singly infected with wCon or Cardinium.
Several strains of Wolbachia, such as wAu from D. simulans and wMel from D. melanogaster, and wStri in the brown planthopper Nilaparvata lugens [15,45], have invaded natural populations even though they induce little to no CI under field conditions. It has been hypothesized that these strains provide an as yet undermined fitness benefit to their hosts. Our results provide important evidence for this hypothesis. The longer-lived multiple-infected females are conducive to the spread of wOri, because they have a greater number of offspring. In addition, the non-CI wOri appears to have a novel mechanism for invading and diffusing through a population. The survival curves of uninfected females indicate that the shorter life span of uninfected females induced by mating with multiple-infected males acts to spread wOri specifically in multipleinfected individuals. This is because the multiple-infected spermatozoa, which decrease the longevity of uninfected females, help to decrease the number of uninfected offspring. We conjecture that the ability of reduction of the longevity of uninfected females probably acts as a new and specific force for wOri to invade and spread through this spider mite population. In addition, fecundity of multiple-infected females affects the survival success of wOri in a similar way. Both the multiple-infected and double-infected males can suppress the fecundity of uninfected females. Although the multiple-infected males are stronger than double-infected males in depressing the fecundity of uninfected females, it is uncertain whether multiple-infection can succeed in the competition with double-infection.
Several studies have proved that the spread of bacterial symbionts can be driven by various host fitness benefits. Dobson et al [14] showed that the combination of CI and host fecundity increase can reduce the threshold infection frequency required for Wolbachia invasion and accelerate the rate of Wolbachia invasion in the super-infected A. albopictus population. In contrast, a higher threshold infection frequency and slower replacement rate is shown in the single-infected A. albopictus population, since it is driven only by CI. In addition, the rapid spread of Rickettsia sp. nr. bellii in a population of an invasive agricultural pest, the sweet potato whitefly, Bemisia tabaci is driven by fitness benefits and female bias [46]. These results support our hypothesis that the spread of wOri can be driven by the combination of CI shared with wCon and Cardinium, host longevity increase and attenuating effects on fecundity and longevity of uninfected females.
In conclusion, we describe novel forces driving the spread of Wolbachia. The new forces we found may provide insights into the distribution of Wolbachia bacteria among species. In addition, the new associations of inherited wOri and hosts that have been regarded as forces driving the spread of the symbiont may result in rapid evolution of both partners and phenotypic shifts that optimize the symbiosis [47]. They can also be exploited to introduce desirable traits in to arthropod hosts using Wolbachia [12,48].